LD59 postmortem: creating an alien language for a 48-hour jam

When I saw that the theme was “signal” for LD59, I immediately knew what I wanted to do: create a game where the player is deciphering a language from an alien signal, melding together two of my biggest interests (sci-fi and linguistics). Because I ultimately shipped a game that felt pretty polished (IMO) but was unfinished puzzle-wise, I felt it might be interesting and/or helpful to write up my experience creating a conlang (CONstructed LANGuage) and a game about it in less than 48 hours.

BACKGROUND/LIMITATIONS

I’ve played around a bit with constructing conlangs before but never with an explicit goal in mind, mostly just because it was fun. Having played games like Chants of Sennaar and Heaven’s Vault before, I already knew a bit about language creation in the game dev space, but let me tell you: it is HARD. There are a few main limitations that are useful to consider when attempting to create a “realistic” alien language:

  1. You have almost no shared contextual reference to work from. Consider the Pioneer plaque, created to adorn the first spacecraft to leave the solar system. In developing the plaque, Carl Sagan found himself extremely limited, as it was not safe to assume that any of our references would be understandable by any intelligent life who found it. For example, arrows are rooted in our own ancient weaponry and may not mean the same thing to anyone else looking at it, no matter how obvious it might seem to us. Even conventions like direction, reading order, orientation, etc. cannot be assumed (they can’t even be assumed within Earth-based languages either, to be fair).
  2. You have almost no shared visual or experiential reference to work from. When trying to learn a language from zero, you pretty much need references. But when you’re trying to learn an ALIEN language from zero, you mostly don’t have them. No gesturing at a tree and then saying the word “tree.” Because, they probably don’t have trees.
  3. You cannot fall back on Earth-based grammatical structures. “Universal grammar” is a linguistic theory that states that all HUMAN languages have a shared underlying structure (again, vastly oversimplifying). That’s not going to quite hold for a language created by aliens many light-years away.

THE LANGUAGE

(From here on there are light spoilers for my game, Adventures in Xenolinguistics. Don’t read unless you have already played/don’t care about spoilers!)

While trying to be true to these limitations, I realized that there was no way I would be able to successfully both create a language AND make it “solvable” by players who aren’t necessary huge puzzle/language dorks like me. Especially not in a 48-hour time frame. So I started with basics. What is the smallest number of types of words with which I can retain communication depth and breadth? I settled on question-based categories and a question-based language. This didn’t come through strongly in the compo game, though I intend to make it more clear for future work on the game. The idea is that every idea can be boiled down into the answer to a question (wait did I accidentally make a Jeopardy language? Oh well.).

Take this sentence:

I walked to the grocery store yesterday.

This could be broken down into:

Who? ME /// What happened? WALKED /// Where? TO /// What? STORE /// Type? FOOD /// When? YESTERDAY

I played around with mixing around the order of these phrases and the meaning was still basically understandable.

This sort of structure lead me to have some ideas about the aliens themselves. With six/seven basic sentence categories, perhaps they themselves had six hyper-expressive limbs that each handled a different element of their language. Maybe they are an inherently very curious culture which lead to their language being designed around questions.

So, I was excited to have come up with this general idea, but the real difficulty came in trying to create the writing system.

My first ideas and sketches centered around a circle-based system that mirrored the aliens’ own anatomy:

Early concept sketch

I quickly realized that I needed to abandon some of the limitations I discussed above. It was just too much to expect a player to understand.

I opted instead for a slightly more human-friendly structure in which the symbols were laid out in a row, and subordinate clauses were arranged above the original symbols:

Once I had a solid idea, I started actually creating symbols for the language. The symbols did evolve a bit as I was creating in-game assets. I "did it live", so to speak, and most of the symbols were only created as I needed them, right into the spritesheet I was using them from. What could go wrong?!

I spent a lot of time making sure that symbolism was used consistently across the language. For example, this is the series of symbols I created to fall under the “where” category:

meanings in rot13: jurer, bhg, hc, ba, gb

When a verb references an object, I used a square shape (what) as opposed to a circular shape (who). I also used an abstraction of the alien’s physical form as their own word for themselves, which was then used throughout the language in various ways:

meanings in rot13: ynfvyyna (bhe anzr sbe gurz), obea, pvivyvmngvba

THE INTENDED PUZZLE STRUCTURE

First off, I knew that this was going to be really hard, so I opted to build out a complete vertical slice of the very beginning of the game before I expanded content. This turned out to be a really good idea, because here’s how the jam schedule went for me:

Friday night: settle on an idea, do research, devise basic structure of language

Saturday morning: create some art, start coding

Saturday all day: code code code

Saturday night: more art while laying in bed

Sunday morning: code code code

Sunday midday: finished vertical slice

Sunday afternoon: bug squash and rush to get in even a little bit more content

The main roadblock for content creation is that it was (and is) incredibly difficult to teach someone a language with no frame of reference, especially when communication is one-way. As I mentioned earlier, no gesturing at a tree and then saying the word “tree.” So the design space is: what would an alien civilization who is trying to communicate with us for an urgent need send to us in order to make sure they’re understood?

Most importantly, I needed the aliens to be able to send a lot of pictures. Luckily (and with a bit of inspiration from the book/film Contact as well as from this puzzle from this year’s MIT Mystery Hunt(link)), I hit upon the idea that the aliens sent a three-part signal that, when properly aligned and translated into three colors, formed images. That’s basically just pixel art!

I assumed that the aliens would have thought about the same dilemma (what would an alien civilization with no shared frame of reference need from us to make sure they understand us?). For me, the answer was in clear image/symbol relationships as well as in repetition. This was the basis for the first couple of image sets intended to teach the user the words related to time and the word for the aliens. I used both images of the aliens’ lifecycle as well as an image of a fire burning through an object to indicate the passage of time:

By having a limited set of base word types, my idea was that if I can teach the player those base types, they’ll have much more ability to make inferences on other words based on their types, especially if those words use shared symbolism. For example, here are a set of words related to thought:

meanings in rot13: yrnea, xabj, fzneg, cyna/vqrn/gubhtug

If the player can get even one of these symbols, that in combination with the base word types should get them pretty far in deciphering the others. Going forward, that’s the sort of approach I will take in designing further puzzles for this game.

CONCLUSION

All that being said, even though the game I shipped was incomplete, I was and am really excited about this idea, and so few games exist in this space that I’m excited to continue working on it. I definitely did not hit my goal of making games like this more accessible (sorry!!!) and there are some inconsistencies and finer points that I'd like to resolve in my conlang that I didn't really figure out until I was thinking about them in retrospect (for example, I'm using the symbol TYPPE as a phrase indicator and TYPE with its own modifiers in ways that aren't clearly delineated.

Overall - when in doubt, your puzzle is probably too hard. This is a lesson I keep learning over and over.

If you’ve read through this whole thing and played the game and are curious, first off, thank you for your interest! As a reward and/or punishment, here’s a version (subject to change, of course) of one of the documents the player is meant to translate later in the game. Enjoy! Or don't!

Source: itch.io archive

Found a bug?
Tell us on Discord